Detecting plant’s diseases in Greenhouse using Deep Learning

Agriculture is considered the main source of economic development in the world. Agriculture is also the main supply of the world’s food and fabrics. Diseases affecting plants in the agriculture process is considered a crisis since it is a threat to the basic human food supply. Early detection of these diseases will save a large amount of the crops. Our proposed approach aims to detect plant’s diseases grown in greenhouses. This is done by monitoring a greenhouse model using an automated intelligent system. The proposed system is used to speed up the plant growth and detect the plant’s diseases. We used tomatoes to test our proposed system. The detected diseases are early blight, late blight, leaf mold, spider mites, target spot, mosaic virus, septoria, bacterial spot, and yellow leaf curl virus. These diseases usually appear on the leaves of the plants and it is hard to differentiate between them by the naked eye. A deep learning library Fast.ai, is used in building a training model from the given dataset of the diseases to get the highest accuracy. The proposed approach achieved 94.8% accuracy in detecting different types of tomato’s diseases. A Web application is developed to track greenhouse’s growth statistics and get notified if there is any disease found on their plant inside the greenhouse.